We present a new method to constrain the dark matter halo density profiles of disk galaxies. Our algorithm employs a Markov Chain Monte Carlo (MCMC) approach to explore the parameter space of a general family of dark matter profiles. We improve upon previous analyses by considering a wider range of halo profiles and by explicitly identifying cases in which the data are insufficient to break the degeneracies between the model parameters. We demonstrate the robustness of our algorithm using artificial data sets and show that reliable estimates of the halo density profile can be obtained from data of comparable quality to those currently available for low surface brightness (LSB) galaxies. We present our results in terms of physical quantities which are constrained by the data, and find that the logarithmic slope of the halo density profile at the radius of the innermost data point of a measured rotation curve can be strongly constrained in LSB ([Vstar/Vobs]max ~ 0.16) galaxies. High surface brightness galaxies ([Vstar/Vobs]max ~ 0.79) require additional information on the mass-to-light ratio of the stellar population - our approach naturally identifies those galaxies for which this is necessary. We apply our method to observed data for the dwarf irregular galaxy DDO 154 and recover a logarithmic halo slope of -0.39 +- 0.11 at a radius of 0.14 kpc. Our analysis validates earlier estimates which were based on the fitting of a limited set of individual halo models, but constitutes a more robust constraint than was possible using other techniques since it marginalises over a wide range of halo profiles.